<<

Tristichopterids (, Tetrapodomorpha) from the Upper -bearing locality of Strud (Belgium, upper ), with phylogenetic and paleobiogeographic considerations Sébastien Olive, Yann Leroy, Edward Daeschler, Jason Downs, S. Ladevèze, Gaël Clément

To cite this version:

Sébastien Olive, Yann Leroy, Edward Daeschler, Jason Downs, S. Ladevèze, et al.. Tristi- chopterids (Sarcopterygii, Tetrapodomorpha) from the Upper Devonian tetrapod-bearing locality of Strud (Belgium, upper Famennian), with phylogenetic and paleobiogeographic considerations. Journal of , Society of , 2020, 40 (1), pp.e1768105. ￿10.1080/02724634.2020.1768105￿. ￿hal-03099746￿

HAL Id: hal-03099746 https://hal.archives-ouvertes.fr/hal-03099746 Submitted on 6 Jan 2021

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Journal of Vertebrate Paleontology: For Review Only

Tristichopterids (Sarcopterygii, Tetrapodomorpha) from the Late Devonian tetrapod-bearing locality of Strud (Belgium, late Famennian), with phylogenetic and paleobiogeographic considerations

Journal: Journal of Vertebrate Paleontology

Manuscript ID Draft

Manuscript Type: Article

Date Submitted by the n/a Author:

Complete List of Authors: Olive, Sébastien; Royal Belgian Institute of Natural Sciences, Paleontology Leroy, Yann; Royal Belgian Institute of Natural Sciences, Paleontology Daeschler, Edward; Academy of Natural Sciences, Department of Vertebrate Paleontology Downs, Jason; Delaware Valley University, Biology; Academy of Natural Sciences, Vertebrate Biology Clément, Gaël; Muséum national d'Histoire naturelle, Département Histoire de la Terre - UMR 5143 du CNRS;

Key Words: tristichopterid, Strud, Famennian, Langlieria, Eusthenodon, Belgium

Society of Vertebrate Paleontology Page 1 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 Tristichopterids (Sarcopterygii, Tetrapodomorpha) from the Late Devonian tetrapod-bearing 4 5 6 locality of Strud (Belgium, late Famennian), with phylogenetic and paleobiogeographic 7 8 considerations 9 10 11 12 1,2,* 1,3,4 2 13 SEBASTIEN OLIVE, YANN LEROY, EDWARD B. DAESCHLER, JASON P. 14 15 DOWNS,2,5 and GAËL CLEMENT,3 16 17 18 19 1OD Earth and , Royal Belgian Institute of Natural Sciences, Rue Vautier 29, 20 21 22 1000 Brussels, Belgium, [email protected]; 23 24 2Academy of Natural Sciences of Drexel University, 1900 Benjamin Franklin Parkway, 25 26 Philadelphia, Pennsylvania 19103, USA, [email protected]; 27 28 3 29 Centre de Recherche en Paléontologie - Paris, UMR CR2P 7207, MNHN/CNRS/SU, CP 38, 30 31 57 rue Cuvier, 75231 Paris Cedex 05, France, [email protected], 32 33 [email protected]; 34 35 4 36 Laboratoire de Paléontologie, Géosciences, CNRS UMR 6118, Université de Rennes 1, 263, 37 38 avenue du Général Leclerc, 35042 Rennes cedex, France; 39 40 5Department of Biology, Delaware Valley University, 700 East Butler Avenue, Doylestown, 41 42 Pennsylvania 18901, U.S.A., [email protected]. 43 44 45 46 47 * Corresponding author 48 49 50 51 52 53 RH: OLIVE ET AL.—TRISTICHOPTERIDS FROM BELGIUM 54 55 56 57 58 59 60

1 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 2 of 51

1 2 3 ABSTRACT—We describe new material of the tristichopterids cf. Langlieria socqueti and 4 5 6 cf. Eusthenodon wangsjoi and other unassignable tetrapodomorph remains from the late 7 8 Famennian locality of Strud, Belgium. Because of recent improvements in our tristichopterid 9 10 knowledge, a new phylogenetic analysis is presented in addition to a paleobiogeographic 11 12 analysis using the Bayesian Binary MCMC statistical method. A formed by (Langlieria 13 14 15 socqueti ((Edenopteron keithcrooki + Hyneria lindae) (Heddleichthys dalgleisiensis 16 17 (Cabonnichthys burnsi (Mandageria fairfaxi + Eusthenodon wangsjoi))))) represents the 18 19 most highly-nested tristichopterids and the biogeographic analysis supports a Western 20 21 22 European origin for it. The biogeographic analysis additionally proposes several 23 24 tristichopterid dispersal events between Euramerica and from the Givetian to 25 26 Famennian, providing a more complex scenario than the one proposed by the Great Devonian 27 28 29 Interchange. 30 31 32 33 INTRODUCTION 34 35 36 37 38 Tristichopteridae is a clade of elongate, aquatic sarcopterygians with an extratemporal 39 40 located between opercular and lateral extrascapular (Ahlberg and Johanson, 41 42 1997; Johanson and Ahlberg, 1997; Snitting, 2008a). Scales are devoid of , rounded 43 44 45 and thin with a small median boss on the internal surface. The caudal is trifurcate and 46 47 diamond-shaped in derived forms (Cloutier and Ahlberg, 1996). 48 49 Tristichopterids are known from the Middle Devonian (latest Givetian; Egerton, 1861; 50 51 52 Young et al., 1992; Swartz, 2012) through the Upper Devonian (Famennian; Jarvik, 1952; 53 54 Clément, 2002; Clément et al., 2009). Their body sizes ranged from 30 centimeters to several 55 56 meters in length. These lobe-finned predatory are quite common in the Devonian 57 58 59 vertebrate assemblages of the world with about twenty described , but only 60

2 Society of Vertebrate Paleontology Page 3 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 Eusthenodon wangsjoi has a cosmopolitan distribution (Laurussia and Gondwana; Ahlberg et 4 5 6 al., 2001; Clément, 2002; Clément and Boisvert, 2006; Clément et al., 2009). Unassignable 7 8 tristichopterid remains are known from Morocco (Lehman, 1977; Lelièvre and Janvier, 1986; 9 10 Lelièvre et al., 1993), from Colombia (Janvier and Maisey, 2010; Olive et al., in review) and 11 12 probably from Iran (Davesne et al., 2015). 13 14 15 Several studies have been done on tristichopterids from Belgium (Lohest, 1888; 16 17 Cloutier and Candilier, 1995; Clément, 2002; Clément et al., 2009; Olive et al., 2015a). 18 19 Eusthenodon wangsjoi and Langlieria socqueti were described from Durnal, Namur Province 20 21 22 (Clément, 2002; Clément et al., 2009); cf. Langlieria from Dison and Modave, Liège 23 24 Province (Clément et al., 2009); and unassignable tristichopterid scales from Becco, Liège 25 26 Province (Olive et al., 2015a). Tristichopterid remains have also been reported from the 27 28 29 locality of Strud, Namur Province (Clément et al., 2009; Denayer et al., 2016:table 1); and 30 31 Evieux, Liège Province (Clément et al., 2009). 32 33 We report here on new tristichopterid material from the tetrapod-bearing locality of 34 35 Strud. We also provide a new phylogenetic analysis that was prompted by recent 36 37 38 improvements in our knowledge of tristichopterid diversity (Swartz, 2012; Schultze and 39 40 Reed, 2012; Young et al., 2013; Borgen and Nakrem, 2016; Daeschler and Downs, 2018; 41 42 Downs et al., 2018). We use our resulting cladogram as the basis for a study of the 43 44 45 paleobiogeographic history of Tristichopteridae. 46 47 48 49 GEOLOGICAL SETTING 50 51 52 53 54 During the late Famennian, Belgium was located along the southeastern margin of 55 56 Euramerica (London-Brabant Peninsula; Ziegler, 1990). In Wallonia (southern Belgium), 57 58 Devonian outcrops are present in the Dinant Synclinorium (allochtonous), Vesdre Area, 59 60

3 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 4 of 51

1 2 3 Theux Tectonic Window, Brabant Parautochton, and Haine-Sambre-Meuse Overturned 4 5 6 Thrust Sheets (HSM OTS; Fig. 1). These Variscan structures form the Namur-Dinant Basin, 7 8 which developed during Devonian and times (Thorez et al., 2006). Famennian 9 10 outcrops from Belgium show a regressive trend from shales formed in an open marine 11 12 environment (Famenne Group, early Famennian) to fluvio-lagunal siltstones, sandstones, and 13 14 15 dolomites (Evieux Formation, lower Famennian) (Thorez and Dreesen, 1986). Late 16 17 Famennian deposits from HSM OTS, Vesdre Area, and Theux Tectonic Window are more 18 19 proximal and the fluvio-lagunal facies are generally better developed than in contemporary 20 21 22 deposits from the Dinant Synclinorium (Thorez et al., 2006). The strata at Strud belong to the 23 24 HSM OTS. Tristichopterid remains were found in beds A, B, D–D’, and E of lithological unit 25 26 7 (Fig. 1). Lithological unit 7 belongs to the late Famennian Evieux Formation. The 27 28 29 sedimentology of this unit corresponds to the filling sequence of a channel in an alluvial plain 30 31 (Denayer et al., 2016). 32 33 34 35 MATERIAL AND METHODS 36 37 38 39 40 Preparation and Imaging 41 42 All specimens were mechanically prepared except for a few specimens that were 43 44 45 virtually prepared by conventional X-ray tomography at the AST-RX (Accès Scientifique à la 46 47 Tomographie à Rayons X) facility of the MNHN. Scanning parameters were as follows: 48 49 voltage = 85 kV, current = 260 µA; for IRSNB P 9918, number of slices = 3000, voxel size = 50 51 52 0.029 mm. IRSNB P 9919 is included in three rock blocks that were scanned independently. 53 54 IRSNB P 9919a, number of slices = 1500, voxel size = 0.098 mm; IRSNB P 9919b, number 55 56 of slices = 2000, voxel size = 0.098 mm; IRSNB P 9919c, number of slices = 1800, voxel 57 58 59 size = 0.116 mm. IRSNB P 9918 and 9919 were virtually reconstructed with MIMICS 60

4 Society of Vertebrate Paleontology Page 5 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 (Materialise’s Interactive Medical Image Control System) 16.0 Software (Materialise Inc.; 4 5 6 proprietary software at the MNHN) at the Atelier 3D, Centre de Recherche en Paléontologie - 7 8 Paris, MNHN. 9 10 11 12 Phylogenetic Analysis 13 14 15 Ahlberg and Johanson (1997) generated a phylogenetic analysis of Tristichopteridae 16 17 that has served as a starting point for subsequent analyses focusing on the clade (e.g. 18 19 Johanson and Ahlberg, 2001; Snitting, 2008b; Clément et al., 2009; Swartz, 2012). The 20 21 22 phylogenetic study developed for the present paper considers 15 species and is the first 23 24 published analysis to include Hyneria lindae and Edenopteron keithcrooki. We chose to 25 26 perform a species-level analysis because we did not want to assume that the genera are 27 28 29 monophyletic. The analysis used 28 morphological characters (SI 1–2). Twenty-three 30 31 characters come directly or are modified from previous analyses, and five new characters 32 33 (23–28) are introduced here. In an effort to minimize qualitative assessments in character 34 35 scoring, we included only characters with well-defined states. We checked character scores 36 37 38 against published descriptions and figures, as well as original material for Hyneria lindae and 39 40 Langlieria socqueti. 41 42 We used PAUP 4.0a162 (Swofford, 2002) and the branch-and-bound search algorithm 43 44 45 to analyze the complete matrix (SI 2). All characters were ordered, except for character #26 46 47 (maximum height of maxilla in external view). We chose andrewsae and 48 49 Spodichthys buetleri for the outgroups because they are well known and phylogenetically 50 51 52 close to tristichopterids (Long, 1985a; Long et al., 2006; Snitting, 2008b; Holland and Long, 53 54 2009). All characters were assigned equal weight and only characters 10, 26, and 28 were 55 56 defined as multistate. Bremer indices and the minimal and maximal number of 57 58 synapomorphies have been produced by PAUP at each node to test the node robustness. 59 60

5 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 6 of 51

1 2 3 Details of the phylogenetic analysis, and notably the character state changings, are available 4 5 6 in SI 3. Only a few character state changings have been reported on the tree in order not to 7 8 overload the figure. 9 10 11 12 Biogeographic Analysis 13 14 15 A biogeographic analysis was performed using Bayesian Binary MCMC (BBM), in 16 17 RASP 3.1 software (Yu et al., 2015). This is a statistical method that uses Bayesian inference 18 19 to reconstruct the ancestral state or biogeographical origin of (Ronquist and 20 21 22 Huelsenbeck, 2003). We applied BBM on the consensus tree resulting from the phylogenetic 23 24 analysis. Nine geographical areas were designated (, Greenland, Canada, 25 26 Pennsylvania, Nevada, Western Europe, Eastern Europe/Russia, South and Antarctica) 27 28 29 and each was assigned to the geographical area(s) where it was found (SI 4). The 30 31 geographic categories were made according to the Late Devonian paleogeography proposed 32 33 by Scotese (2014) and Ziegler (1990). For instance, , England, Belgium were all 34 35 parts of the Rhenish Basin during the Late Devonian (Ziegler, 1990) and thus were grouped 36 37 38 together as “Western Europe”. The BBM analysis was performed over 106 generations with 39 40 10 Markov chains, sampling every 100 generations, with the temperature set to 0.1. The first 41 42 one hundred trees were discarded as part of the burn-in period, and the Jukes-Cantor model 43 44 45 was used, with equal among-site rate variation. The results of the biogeographic 46 47 reconstructions for each node are presented in SI 5. The results of the analysis are expressed 48 49 as the probability of a given clade originating from each of nine continental areas. 50 51 52 53 54 Abbreviations 55 56 Institutional Abbreviations—IRSNB, Institut royal des Sciences naturelles de 57 58 Belgique, Brussels, Belgium; MNHN, Muséum national d’Histoire naturelle, Paris, France. 59 60

6 Society of Vertebrate Paleontology Page 7 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 Anatomical Abbreviations—a.ad, anterior wall of adductor fossa; ar.hyo, 4 5 6 articulation area with hyomandibular; ar.pal, area for palatoquadrate dorsal process 7 8 (paratemporal process); aut.p, autopalatine process; bu, bulge; cl.d, postero-dorsal 9 10 crest; cl.m, cleithrum mesial crest; co.1, anterior coronoid; co.2, middle coronoid; co.3, 11 12 posterior coronoid; d, dentary; d.d, dentary marginal dentition; d.f, dentary fang; d.s, dentary 13 14 15 suture; ent.pa, anterior process of entopterygoid; f.co.1, anterior coronoid fang; f.co.2, 16 17 middle coronoid fang; f.co.3, posterior coronoid fang; f1.co.3, anterior fang of posterior 18 19 coronoid; f2.co.3, second fang of posterior coronoid; f3.co.3, posterior fang of posterior 20 21 22 coronoid; fo.ico, intercoronoid fossa; fo.pco, precoronoid fossa; inf.1-4, infradentaries; l.d, 23 24 dorsal lamina of cleithrum; l.pb, dorsal blade post-branchial lamina of cleithrum; l.v, ventral 25 26 lamina of cleithrum; la.l, lateral vertical lamina; la.m, mesial vertical lamina; m.ad, posterior 27 28 29 part forming the mesial part of the adductor fossa; md.g, mandibular sensory groove; o.r, 30 31 oblique ridge; ov.ano, area of cleithrum overlapping anocleithrum; ov.br, branchial 32 33 overlapping area; ov.cla, clavicle overlapping area; ov.it, intertemporal 34 35 overlapping area; ov.j, jugal overlapping area; ov.l.ext, lateral extrascapular overlapping 36 37 38 area; ov.mx, maxilla overlapping area; ov.p.sr, posterior supraorbital overlapping area; 39 40 ov.po, postorbital overlapping area; ov.pop, preopercular overlapping area; ov.psp, 41 42 postspiracular overlapping area; ov.qj, quadratojugal overlapping area; ov.sm1, 43 44 45 submandibular 1 overlapping area; ov.sq, squamosal overlapping area; ov.st, supratemporal 46 47 overlapping area; ov.ta, tabular overlapping area; p.g, parietal sensory groove; pi, pineal 48 49 area; pr.a, anterior process; pr.ad, antero-dorsal process; pop.g, preopercular sensory 50 51 52 groove; ppr, posterior process of postorbital; r.f, replacement pit of middle coronoid fang; 53 54 ri.it.so, ridge dividing intertemporal and postorbital overlapping areas; ri.ioc, ridge housing 55 56 infraorbital sensory canal; sp.n, spiracular notch. 57 58 59 60

7 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 8 of 51

1 2 3 SYSTEMATIC PALEONTOLOGY 4 5 6 7 8 SARCOPTERYGII Romer, 1955 9 10 Berg, 1937 11 12 TRISTICHOPTERIDAE Cope, 1887 13 14 15 cf. LANGLIERIA Clément, Snitting and Ahlberg, 2009 16 17 18 19 Type Species—Langlieria socqueti Clément, Snitting and Ahlberg, 2009. 20 21 22 23 24 cf. LANGLIERIA SOCQUETI Clément, Snitting and Ahlberg, 2009 25 26 (Fig. 2) 27 28 29 30 31 Referred Material—IRSNB P 9912, partial left squamosal (bone has been 32 33 mechanically removed to have access to the mould of the external surface), and IRSNB P 34 35 9913, right . 36 37 38 Locality and Horizon—Strud quarry, Namur Province, Belgium, Evieux Formation, 39 40 late Famennian, Late Devonian. 41 42 Remarks—Two tristichopterid species have been described from Belgium: 43 44 45 Eusthenodon wangsjoi and Langlieria socqueti. We assign the following material (isolated 46 47 squamosal and lower jaw) to cf. Langlieria socqueti according to dermal ornament and size 48 49 category. Of the two specimens described here, only one (IRSNB P 9913) exhibits a feature 50 51 52 included in the diagnostic character combination of Clément et al. (2009) and this feature 53 54 (presence of dentary fang pair) is present in most tristichopterids. The one autapomorphy that 55 56 Clément et al. (2009) cite for L. socqueti (absence of marginal vomerine teeth; Clément et al., 57 58 2009) is impossible to address in the new material. 59 60

8 Society of Vertebrate Paleontology Page 9 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 4 5 6 Description 7 8 Squamosal—The squamosal (IRSNB P 9912; Fig. 2A–C) exhibits a prominent 9 10 rostrodorsal process and a less prominent rostromedial process. The dorsal margin of the 11 12 bone is concave and the dorsocaudal corner is rounded and located well rostral to the 13 14 15 caudoventral portion of the squamosal. The dermal surface of the squamosal contains wide, 16 17 conjoined areas for the overlap of the postorbital and jugal bones. There is a narrow area for 18 19 the overlapping maxilla along the ventral margin. The relationship between squamosal and 20 21 22 maxilla is reversed in IRSNB P 9912 relative to the condition in a partial of L. socqueti 23 24 (MNHN ARD 250) from the type locality (squamosal overlaps maxilla; Clément, 2002:text- 25 26 figs. 8–9). Maxilla overlap onto the squamosal is the more common condition in 27 28 29 Tristichopteridae; the reversed condition is otherwise known only in Eusthenodon wangsjoi 30 31 and possibly Platycephalichthys bischoffi (Clément, 2002). On the visceral surface of IRSNB 32 33 P 9912, along the dorsal margin, there is an elongated and narrow area of overlap onto 34 35 supratemporal, tabular, and possibly postspiracular. 36 37 38 Lower Jaw—The dentary displays a large fang pair with cutting edges and ridges on 39 40 its base. Ridges are absent or not preserved on the apex. As in MNHN ARD 251 (Clément et 41 42 al. 2009:text-fig. 7), there is no marginal dentition of the dentary labial to the fangs or 43 44 45 between the fangs and the mandibular symphysis (IRSNB P 9913, Fig. 2D–F). Among 46 47 tristichopterids, a marginal dentary tooth row that appears only distal to the fangs is 48 49 additionally observed in Cabonnichthys burnsi (Johanson and Ahlberg, 1997), Edenopteron 50 51 52 keithcrooki (Young et al., 2013), and Eusthenodon wangsjoi (Jarvik, 1952; Johanson, 2004). 53 54 The distal end of infradentary 4 in IRSNB P 9913 shows a wide and ventrally slanting 55 56 area of overlap for the quadratojugal. Ventral to the quadratojugal overlap area is a long 57 58 59 60

9 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 10 of 51

1 2 3 eroded mandibular groove. The overlapping area for the first submandibular can be observed 4 5 6 on the ventral part of infradentary 1 (IRSNB P 9913, Fig. 2D–E). 7 8 Only one fang of coronoid 3 is visible in IRSNB P 9913 (Fig. 2D–F). The number of 9 10 fang pairs on coronoid 3 is impossible to determine in this specimen and the condition is 11 12 unknown in L. socqueti. Among tristichopterids, only Spodichthys buetleri (Snitting, 2008b), 13 14 15 Tristichopterus alatus (Snitting, 2009), and E. keithcrooki (Young et al., 2013) are reported 16 17 with one fang pair, rather than two, on coronoid 3. The coronoid fangs of IRSNB P 9913 18 19 exhibit cutting-edges and striations at the base that may or may not reach to the apex of the 20 21 22 tooth. 23 24 Dermatocranial Ornament—The dermatocranial ornament of IRSNB P 9912 and 25 26 IRSNB P 9913 is very fine. IRSNB P 9912 (squamosal) exhibits subparallel ridges in the 27 28 29 rostrodorsal part of the bone and anastomosing ridges elsewhere. On IRSNB P 9913 (dentary 30 31 + infradentaries), the ornament is mainly made of smooth anastomosed ridges, which are 32 33 more strongly pronounced at the distal end of the dentary. 34 35 36 37 38 cf. EUSTHENODON Jarvik, 1952 39 40 41 42 Type Species—Eusthenodon wangsjoi Jarvik, 1952. 43 44 45 46 47 cf. EUSTHENODON WANGSJOI Jarvik, 1952 48 49 (Figs 3–4, SI 6–7) 50 51 52 53 54 Referred Material—IRSNB P 9910, partial left parietal; IRSNB P 9914, left 55 56 supratemporal, part and counterpart; IRSNB P 9915, partial left postorbital; IRSNB P 9916, 57 58 59 60

10 Society of Vertebrate Paleontology Page 11 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 left preopercular; IRSNB P 9917, left opercular; IRSNB P 9918, right dentary; IRSNB P 4 5 6 9919, partial right mandible. 7 8 Locality and Horizon— Strud quarry, Namur Province, Belgium, Evieux Formation, 9 10 late Famennian, Late Devonian. 11 12 Remarks—These specimens are referred to cf. Eusthenodon wangsjoi on account of 13 14 15 dermal ornament and size category. In addition, IRSNB P 9915 (postorbital) shares general 16 17 shape characteristics with the E. wangsjoi holotype (P. 1476) in addition to a well-developed 18 19 rostrodorsal process (Jarvik, 1952:pl. 9). IRSNB P 9917 (preopercular) has a wide 20 21 22 rostroventral process similar to that observed in the E. wangsjoi holotype (Jarvik, 1952:pl. 23 24 10). General shape characteristics are shared between IRSNB P 9914 (supratemporal) and 25 26 IRSNB P 9918 (opercular) and the E. wangsjoi holotype (Jarvik, 1952:pl. 9). IRSNB P 9919 27 28 29 (mandible) shares with the E. wangsjoi holotype (Jarvik, 1952:pl. 10) a lack of fusion among 30 31 the dermatocranial mandibular bones. 32 33 34 35 Description 36 37 38 Parietal—IRSNB P 9910 (Fig. 3A–B) is an isolated partial parietal that preserves 39 40 much of the caudal margin and the medial margin. The dermal surface of the parietal exhibits 41 42 a short area of supratemporal overlap along the caudolateral margin. The posterior oblique 43 44 45 pitline groove is observed lateral to the pineal region. The shape of the parietal suggests a 46 47 kite-shaped pineal series, contrary to the tear-drop shaped condition in the Eusthenodon 48 49 wangsjoi holotype (P. 1476; Jarvik, 1952:pl. 9) and in Clément’s (2002) emended diagnosis 50 51 52 of the species. In addition, it appears as though the pineal series either reaches or nearly 53 54 reaches the caudal margin of the parietal shield. Several specimens of E. wangsjoi (including 55 56 the holotype, P. 1476; Jarvik, 1952:pl. 9) show a more rostrally positioned pineal series. High 57 58 59 60

11 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 12 of 51

1 2 3 variability of pineal positioning in E. wangsjoi has previously been recognized by Jarvik 4 5 6 (1985). 7 8 Supratemporal—IRSNB P 9914 (Fig. 3C–F) is widest along its caudal margin and 9 10 the spiracular notch is strongly hook-shaped. This shape is quite different from what is 11 12 observed in the Eusthenodon wangsjoi holotype (P. 1476; Jarvik, 1952) and more closely 13 14 15 resembles the condition observed in foordi (e.g., Jarvik, 1944a). Areas 16 17 overlapped by the intertemporal and postorbital are observed on the dermal surface of the 18 19 supratemporal (Fig. 3C), in the rostrolateral corner. A strong ridge divides the two overlap 20 21 22 areas (Fig. 3E) as in Langlieria socqueti (Clément, 2002:text-fig. 6B). On the visceral 23 24 surface, a smooth depression is visible anteriorly. This depression may have received the 25 26 palatoquadrate paratemporal process (as proposed by Clément, 2002, for E. wangsjoi). The 27 28 29 ridge housing the infraorbital canal is present around this depression. A possible articulation 30 31 area with the is located in the caudomedial corner (Fig. 3F). 32 33 Postorbital—IRSNB P 9915 (Fig. 3G–H) is an isolated partial postorbital with an 34 35 extended shared area for the overlap of both intertemporal and posterior supraorbital bones. 36 37 38 These overlap areas are not observable in the Eusthenodon wangsjoi type material. 39 40 Preopercular—IRSNB P 9916 (Fig. 3I–J) is a partial left preopercular that preserves 41 42 an area for squamosal overlap that appears short but may be incomplete. 43 44 45 Opercular—IRSNB P 9917 (Fig. 3K–L) is an opercular with a high height/length 46 47 ratio (1.5). Jarvik (1952) described the opercular of Eusthenodon wangsjoi as higher than 48 49 long and figures a single specimen (P. 1481, Jarvik, 1952:pl. 12:2) with a ratio that is not as 50 51 52 high (=1.24) as that in IRSNB P 9917. The ventral margin of IRSNB P 9917 is rounded as in 53 54 Eusthenopteron foordi (e.g., Jarvik, 1944a) and is notably unlike the condition in 55 56 Cabonnichthys burnsi (Ahlberg and Johanson, 1997:fig. 3), where the margin is indented by 57 58 the subopercular. The dorsal margin of IRSNB P 9917 shows a well-developed and large 59 60

12 Society of Vertebrate Paleontology Page 13 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 overlap area divided in two parts by a ridge. The larger rostral part was probably for the 4 5 6 postspiracular bone and the smaller caudal part for the lateral extrascapular. In E. wangsjoi, 7 8 only the overlap area for the postspiracular has been previously reported (by Jarvik, 1952:pl. 9 10 12:2). However, Jarvik (1952:64) expresses doubt over whether the area was overlapped by 11 12 the postspiracular or the lateral extrascapular. The rostral margin of IRSNB P 9917 displays a 13 14 15 very narrow overlap area for the preopercular. 16 17 Lower Jaw—IRSNB P 9918 (Fig. 4A–B) exhibits a dentary fang pair and no 18 19 marginal teeth are observed from the fang pair to the symphysis. Within Tristichopteridae, 20 21 22 this condition is shared among Cabonnichthys burnsi (Johanson and Ahlberg, 1997), 23 24 Edenopteron keithcrooki (Young et al., 2013), and Eusthenodon wangsjoi (Jarvik, 1952; 25 26 Johanson, 2004). IRSNB P 9919 (Fig. 4D–F) includes all three coronoids. Coronoid 3 is 27 28 29 much longer than coronoids 1 or 2. Among tristichopterids, only Spodichthys buetleri 30 31 (Snitting, 2009) and Tristichopterus alatus (Snitting, 2009) have been reported with a third 32 33 coronoid of a length subequal to coronoids 1 and 2. The fangs of coronoids 1 and 2 have 34 35 sharp carinae along mesial and distal margins. The fangs of both pairs on coronoid 3 are 36 37 38 much smaller than those of coronoids 1 or 2. No striations are visible on the coronoid fangs 39 40 of IRSNB P 9919. A marginal tooth row is not visible on any of the three coronoids of 41 42 IRSNB P 9919. This may be a result of preservation; no other tristichopterid is without 43 44 45 marginal coronoid dentition. Among tristichopterids, the most reduced condition previously 46 47 recognized (marginal teeth on third coronoid only) is observed in E. wangsjoi (Jarvik, 1952), 48 49 Hyneria lindae (Daeschler and Downs, 2018), and M. fairfaxi (Johanson and Ahlberg, 1997). 50 51 52 Dermatocranial Ornament—As described in Clément’s (2002) emended diagnosis 53 54 for Eusthenodon wansgjoi, the dermatocranial ornament of these Strud specimens consists of 55 56 ridges forming distinct networks and independent tubercles are rare. 57 58 59 60

13 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 14 of 51

1 2 3 DESCRIPTION OF OTHER TRISTICHOPTERID MATERIAL FROM STRUD 4 5 6 7 8 Unless otherwise noted, the specimens reported below are referred to cf. 9 10 Tristichopteridae sp. indet. according to shapes, sizes, and dermal ornaments that compare 11 12 13 more favourably to Strud’s tristichopterids than to the other vertebrates at the locality 14 15 (acanthodians, actinopterygians, antiarchs, arthrodires, phyllolepids, porolepiforms, 16 17 rhizodonts, osteolepidids, dipnoans, ; Lohest, 1888; Leriche, 1931; Clément et al., 18 19 2004; Clément and Boisvert; 2006; Olive, 2015; Olive et al., 2015b, 2016a, 2016b). Absence 20 21 22 of diagnostic features prevents more specific taxonomic assignments. Taxonomic caution is 23 24 especially warranted in the cases of isolated tetrapodomoprh cleithra and suboperculars. 25 26 Several authors have reported on the intraspecific variability of cleithrum morphology and 27 28 29 ornament and of subopercular morphology in tetrapodomorphs (Andrews and Westoll, 1970; 30 31 Jarvik, 1944b). For many of the specimens presented here, very little meaningful anatomical 32 33 description is possible. Our intention then is to simply present an inventory of potential 34 35 36 tristichopterid material produced by recent collecting efforts at the Strud locality. 37 38 IRSNB P 9920 (Fig. 5A) is a right preopercular with a deep dorsal–ventral sensory 39 40 groove in its caudal half. Its ornamentation comprises small tubercles and ridges. 41 42 IRSNB P 9921–9923 (Fig. 5B–D) are isolated subopercular bones. The rostral process 43 44 45 of IRSNB P 9921 is especially long and rod-like (Fig. 5B). IRSNB P 9923 is much smaller 46 47 than the other two and has a more rounded shape with a short, stout rostral process (Fig. 5D). 48 49 IRSNB P 9911 (Fig. 5E–F) is a partial right entopterygoid and palatoquadrate. Much 50 51 52 of the preserved entopterygoid is without denticles although this is possibly due to 53 54 weathering. A large and rounded oblique ridge is present in the middle of the bone. This 55 56 ridge runs caudodorsally and broadens from a prominent rounded bulge (Fig. 5F). The rostral 57 58 59 end of the specimen exhibits a pronounced autopalatine process and a ventral process of the 60

14 Society of Vertebrate Paleontology Page 15 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 entopterygoid with a short triangular groove between them that tapers caudally. A prominent 4 5 6 ascending process is partially observed at the dorsal preservational limit. A row of large and 7 8 pointed denticles appear along the ventral margin of the entopterygoid’s rostral end. The 9 10 ventral margin of the palatoquadrate is relatively straight and does not exhibit the wide lateral 11 12 flange that surrounds the caudal end of the ectopterygoid in Eusthenodon wangsjoi (Jarvik, 13 14 15 1952:pl. 16:2). 16 17 IRSNB P 9924–9927 are isolated cleithra. The dorsal lamina ornament of IRSNB P 18 19 9924 (Fig. 6A–B) is a network of generally parallel fine anastomosed crests; that of IRSNB P 20 21 22 9926 and IRSNB P 9927 (Fig. 6E–F) is coarse tubercles and small ridges similar to cleithrum 23 24 ornament of Platycephalichthys bischoffi (Vorobyeva, 1959, 1962) and Langlieria socqueti 25 26 (Clément, 2002; Clément et al., 2009). The dorsal lamina of IRSNB P 9925 (Fig. 6D) is 27 28 29 generally smooth, perhaps due to weathering. In IRSNB P 9925, a long crest follows the 30 31 caudal margin of the dorsal lamina’s visceral surface. (Fig. 6D). 32 33 IRSNB P 9928 and 9929 (Fig. 6G–H) are isolated fangs with ridges along their base 34 35 and sharp carinae along mesial and distal edges. 36 37 38 IRSNB P 9930–9933 (Fig. 6I–L) are isolated scales. Each shows the visceral boss that 39 40 allows an assignment to Tristichopteridae sp. indet. Several (IRSNB P 9930–9931, IRSNB P 41 42 9933) show radiating grooves on the free margin. 43 44 45 46 47 RESULTS 48 49 50 51 52 Phylogenetic Analysis 53 54 The phylogenetic analysis of the complete data matrix yielded three most 55 56 parsimonious trees with a score of 62 (CI = 0.5; RI = 0.617; RCI = 0.309). Figure 7 presents 57 58 59 the 50% majority-rule consensus tree that was generated from the data. Bremer indices are all 60

15 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 16 of 51

1 2 3 equal to 1 except the clade (Cabonnichthys burnsi (Mandageria fairfaxi + Eusthenodon 4 5 6 wangsjoi)), which has a Bremer index of 2. 7 8 Usually considered as a tristichopterid, Tristichopterus alatus has an unresolved basal 9 10 position in our phylogenetic analysis. It belongs to the basal polytomy with the two outgroups 11 12 and the rest of the tristichopterids (node 11). 13 14 15 Node 11 is characterized by five synapomorphies: presence of epichordal radials in 16 17 the caudal fin (character 1, state 1), a pineal foramen well posterior to orbits (9,1), two fang 18 19 positions on posterior coronoid (12,1), a posterior coronoid much longer than middle 20 21 22 coronoid (13,1) and a dentary fang pair (18,1). 23 24 Node 10 is characterized by a symmetrical caudal fin (2,1), a posterior process of the 25 26 posterior supraorbital much longer than its orbital margin (7,1), two fang positions on 27 28 29 ectopterygoid (11,1), less than half of the lacrimal rostral to orbit (27,1) and the presence of 30 31 an extratemporal in a postspiracular position (21,1). 32 33 Eusthenopteron (node 1) is characterized by the lack of epichordal radials in the 34 35 caudal fin (1,0), the absence of a dentary fang pair (18,0) and a straight or slightly concave 36 37 38 lateral margin of the vomer’s caudal process (19,1). 39 40 In our analysis, and contrary to Borgen and Nakrem (2016), the ancestral character 41 42 state for the premaxillary dentition (character 10) is not homodonty (morphotype A, state 0), 43 44 45 but instead a gradual tooth size increase with proximity to the symphysis (state 1). 46 47 Homodonty is a derived character, which occurred independently in Spodichthys buetleri, 48 49 Eusthenopteron wenjukowi, and Platycephalichthys bischoffi. A single, greatly enlarged 50 51 52 premaxillary fang (state 2; morphotypes C and D of Borgen and Nakrem, 2016) supports the 53 54 clade of Famennian tristichopterids (node 7). It also appears independently in Tinirau 55 56 clackae. The Famennian tristichopterid clade (node 7) is also supported by the lack of dentary 57 58 marginal teeth near the symphysis (14,1). 59 60

16 Society of Vertebrate Paleontology Page 17 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 In this analysis, the best supported clades are (Hyneria lindae + Edenopteron 4 5 6 keithcrooki) (minimal and maximal number of synapomorphies of five and six) and 7 8 ((Eusthenodon wangsjoi + Mandageria fairfaxi) Cabonnichthys burnsi) (Bremer index of 2). 9 10 The first is characterized by the absence of contact between the posterior supraorbital and the 11 12 intertemporal (5.1), an apical fossa divided in two (20,1), a flat to concave denticulated field 13 14 15 of the parasphenoid (22,0), a 0.5 or greater width/length ratio of vomers + parasphenoid 16 17 (23,0), a 0.5 or greater width/length ratio of parietals (24,1), and a maxilla overlap onto 18 19 squamosal (25,0). Edenopteron keithcrooki has only one fang position on its posterior 20 21 22 coronoids (12,0) and this reversal makes it unique within the Famennian tristichopterid clade 23 24 (node 7). 25 26 The second of the best supported clades, ((Eusthenodon wangsjoi + Mandageria 27 28 29 fairfaxi) Cabonnichthys burnsi), is characterized by the absence of a vertical blade-like 30 31 lamina lateral to the fangs of coronoid 1 (15,1), a marginal coronoid tooth row on coronoid 3 32 33 only (28,2), the presence of accessory vomers (16,1) and a maxilla of equal height along its 34 35 entire length (26,2). E. wangsjoi and M. fairfaxi share the following combination of 36 37 38 characters: jugal excluded from orbital margin (3,1), an absence of contact between posterior 39 40 supraorbital and intertemporal (5,1), a contact between posterior supraorbital and lacrimal 41 42 (6,1), absence of a marginal tooth row mesial to the dermatopalatine fang (17,1) and half or 43 44 45 more of the lacrimal appears rostral to the orbit (27,0). The relationships among E. wangsjoi, 46 47 M. fairfaxi and C. burnsi agree with the results of Ahlberg and Johanson (1997, 1998, 2001) 48 49 and place Cabonnichthys as the sister taxon of (Eusthenodon + Mandageria). 50 51 52 53 54 Biogeographic Analysis 55 56 Clade 7 includes the most derived tristichopterids: Langlieria socqueti, Edenopteron 57 58 keithcrooki, Hyneria lindae, Heddleichthys dalgleisiensis, Cabonnichthys burnsi, 59 60

17 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 18 of 51

1 2 3 Mandageria fairfaxi and Eusthenodon wangsjoi. This node is reconstructed with a Western 4 5 6 Europe origin (~84% probability). 7 8 Clade 4 gathers taxa (Cabonnichthys burnsi, Mandageria fairfaxi and Eusthenodon 9 10 wangsjoi) with an Australian origin (~81% probability) and clade 2 groups taxa 11 12 (Edenopteron keithcrooki and Hyneria lindae) with a possible Australian origin (~67%). 13 14 15 The major conflict in the biogeographic reconstruction occurs at the root node 12 16 17 where the Australian origin is supported at 30%, the Greenland origin at 28%, the Western 18 19 Europe origin at 22% and the Nevada origin at 7%. This is explained by the basal polytomy 20 21 22 and the choice of the outgroups, which are themselves of Australian (Gogonasus andrewsae) 23 24 and Greenland (Spodichthys buetleri) origins. 25 26 The basal position of Tinirau clackae and Notorhizodon mackelveyi suggest a 27 28 29 Western North America (Nevada) origin for the clade 11 (~62% probability) and an Antarctic 30 31 origin for clade 10 (~63% probability). 32 33 34 35 DISCUSSION 36 37 38 39 40 Phylogenetic Analysis 41 42 As mentioned by Schultze and Reed (2012), interrelationships of tristichopterids have 43 44 45 been very stable over the and have changed only by addition of new taxa. This is the 46 47 case in the present analysis. Most recent phylogenetic analyses of tristichopterids support a 48 49 highly nested clade of tristichopterids including variably Cabonnichthys, Eusthenodon, 50 51 52 Mandageria, (Ahlberg and Johanson, 1997, 1998; Johanson and Ahlberg, 2001; Zhu and 53 54 Ahlberg, 2004; Snitting, 2008b; Clément et al., 2009; Swartz, 2012) and Langlieria (Clément 55 56 et al., 2009). The present analysis gathers the same taxa (node 7) and additionally includes 57 58 Hyneria lindae and Edenopteron keithcrooki, both included for the first time in a published 59 60

18 Society of Vertebrate Paleontology Page 19 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 phylogenetic analysis. The clade of Famennian tristichopterids that originates at node 7 4 5 6 additionally includes Heddleichthys dalgleisiensis. 7 8 Position of Spodichthys—Spodichthys is basal to all tristichopterids in all 9 10 phylogenetic analyses where it is included (Snitting, 2008a, 2008b; Clément et al., 2009). It 11 12 displays a number of plesiomorphic characters such as the absence of contact between the 13 14 15 posterior supraorbital and the intertemporal (5,1), the pineal foramen located at the level with 16 17 or anterior to the posterior margin of the orbit (9,0), only one fang position on the 18 19 ectopterygoid and third coronoid (11,0 and 12,0), a third coronoid similar in length to the 20 21 22 middle coronoid (13,0), the presence of a vertical lamina lateral to the coronoid fang on the 23 24 anterior coronoid (15,0), the absence of a dentary fang pair (18,0) and a width/length ratio of 25 26 vomers + parasphenoid equal to or more than 0.5 (23,0). In our analysis, Spodichthys buetleri 27 28 29 appears in a basal polytomy together with Gogonasus andrewsae, Tristichopterus alatus and 30 31 a clade that contains all other Tristichopteridae. The phylogenetic position of S. buetleri 32 33 remains an open question because more than half of the characters are missing data (19 of 34 35 28). It is considered to be the sister group of Tristichopteridae by Snitting (2008b) because it 36 37 38 cannot be demonstrated that its extratemporal reaches into the postspiracular area, a feature 39 40 that diagnoses Tristichopteridae. 41 42 Monophyly of Tristichopteridae—Tristichopterid monophyly has been established 43 44 45 in several studies (Long, 1985b; Ahlberg and Johanson, 1997, 1998; Johanson and Ahlberg, 46 47 2001; Zhu and Ahlberg, 2004; Snitting, 2008a, 2008b; Clément et al., 2009). In the present 48 49 analysis, because of the low resolution at the base of the tree, this monophyly is not 50 51 52 specifically supported. The presence of an extratemporal in a postspiracular position (21,1), a 53 54 feature used to diagnose members of Tristichopteridae, is here regarded as a synapomorphy 55 56 of clade 10 only. It is absent in Tinirau clackae (the ‘postspiracular’ is in a rostral position) 57 58 and its presence in Tristichopterus alatus is regarded as a convergence. 59 60

19 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 20 of 51

1 2 3 Newly Added Taxa—Edenopteron keithcrooki and Hyneria lindae are included in a 4 5 6 phylogenetic analysis for the first time and these appear as sister species to one another (node 7 8 2). Node 2 is supported by six synapomorphies (see Results) including the of the 9 10 apical fossa into two parts, a condition that is unique within Tristichopteridae. 11 12 The position of H. lindae within tristichopterids appears as less derived than 13 14 15 previously thought by several authors (Daeschler and Shubin, 2007; Clément et al., 2009; 16 17 Daeschler and Downs, 2018). In a previous publication by two of us (Daeschler and Downs, 18 19 2018), a highly nested phylogenetic position for H. lindae was proposed due to a suite of 20 21 22 characters that includes contact between parietal and postorbital, a postorbital that does not 23 24 contribute to the orbital margin, a wide and concave denticulated plate of the parasphenoid, a 25 26 dentary fang pair, and a lack of marginal dentition on all but the third coronoid. Except for 27 28 29 the last character (28,2), the new phylogenetic analysis supports all of these as either derived 30 31 features appearing quite early in the history of tristichopterids (dentary fang pair (18,1) and 32 33 lack of postorbital contribution to orbital margin (4,1)) or plesiomorphic for the group (wide 34 35 and concave denticulated plate of the parasphenoid, 22,0). 36 37 38 Edenopteron keithcrooki was also presumed to be phylogenetically highly nested 39 40 (Young et al., 2013) on account of characters including accessory vomers and contact 41 42 between parietal and postorbital. According to our analysis, while the presence of accessory 43 44 45 vomers suggests a derived tristichopterid, the rest of the character states support its sister 46 47 group relationship with H. lindae. 48 49 Tinirau clackae was previously included in the phylogenetic analysis of Swartz 50 51 52 (2012) but this analysis used a large dataset including Devonian and Carboniferous 53 54 tetrapodomorphs. As noticed by Young et al. (2013), the character list diluted characters 55 56 relevant to tristichopterids and pushed T. clackae and Platycephalichthys out of 57 58 Tristichopteridae. In Swartz’s (2012) cladogram, Platycephalichthys was sister group to 59 60

20 Society of Vertebrate Paleontology Page 21 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 Elpistostegalia and Tinirau was sister group to (Platycephalichthys + Elpistostegalia). In our 4 5 6 analysis, the only species in these two groups occupy basal position within Tristichopteridae. 7 8 Mandageriidae—The subfamily Mandageriidae was established by Young (2008) 9 10 and includes Gondwana tristichopterid taxa: Mandageria fairfaxi, Cabonnichthys burnsi, 11 12 possibly Notorhizodon mackelveyi and Edenopteron keithcrooki (referred to as the 13 14 15 “undescribed tristichopterid from near Eden, NSW” in Young, 2008). According to Young 16 17 (2008), who did not perform a phylogenetic analysis to support his assertion, features 18 19 justifying the erection of this subfamily were a shared morphology and the presence of 20 21 22 accessory vomers. This pairing is counter to the relationships presented in Ahlberg and 23 24 Johanson’s (1997) phylogenetic analysis, wherein M. fairfaxi is the sister group of 25 26 Eusthenodon wangsjoi, not Cabonnichthys burnsi. One of Young’s (2008) arguments against 27 28 29 the pairing of M. fairfaxi and E. wangsjoi was the weakness of the only synapomorphy that 30 31 supported it. According to Young (2008), absence of contact between intertemporal and 32 33 posterior supraorbital is impossible to assess in M. fairfaxi. 34 35 In our phylogenetic analysis, we respond to the issues raised by Young (2008). The 36 37 38 “accessory vomers” character has been added (ch.16; as it was already done by Clément et 39 40 al., 2009). We scored character 5 (contact between posterior supraorbital and intertemporal) 41 42 in the same way as Ahlberg and Johanson (1997) for Mandageria fairfaxi and Cabonnichthys 43 44 45 burnsi since we do not have access to the type material, but whereas the absence of contact 46 47 was only a shared character by E. wangsjoi and M. fairfaxi, in our matrix, it is additionally 48 49 shared by E. keithcrooki and H. lindae. We also add characters and key species that should 50 51 52 help to address the validity of the subfamily Mandageriidae (i.e., N. mackelveyi and E. 53 54 keithcrooki). 55 56 The relationships supported by our analysis are those originally presented by Ahlberg 57 58 and Johanson (1997, 1998) and also Clément et al. (2009): (C. burnsi (M. fairfaxi + E. 59 60

21 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 22 of 51

1 2 3 wangsjoi)). The clade (M. fairfaxi + E. wangsjoi) is supported by five synapomorphies and 4 5 6 displays a minimal and maximal numbers of synapomorphies of four and six. Mandageriidae 7 8 appears as paraphyletic here and is therefore not considered to be valid. 9 10 Newly Added Characters—Compared to previous phylogenetic studies on 11 12 tristichopterids, six new characters have been added in our analysis, some of them based on 13 14 15 observations made on Hyneria lindae. Half of the new characters (23, 24, and 25) represent 16 17 unexpected synapomorphies for the clade including Hyneria lindae and Edenopteron 18 19 keithcrooki. Characters 23 and 24 are width/length ratios implying a widening of the palate 20 21 22 and of the , respectively, in clade 17. A marginal coronoid tooth row that is only 23 24 present on coronoid 3 represents a new synapomorphy for the highly nested group of 25 26 tristichopterids: (Cabonnichthys burnsi (Mandageria fairfaxi + Eusthenodon wangsjoi)). 27 28 29 30 31 Biogeographic History of Tristichopteridae 32 33 There is a perfect congruence between the phylogenetic and stratigraphic data (Fig. 34 35 7). The origin of the whole group would be, at least, Givetian in age if we consider the oldest 36 37 38 occurrence of tristichopterids, i.e. Tristichopterus alatus, Notorhizodon mackelveyi, and 39 40 Tinirau clackae (Egerton, 1861; Young et al., 1992; Swartz, 2012; Schultze and Reed, 2012). 41 42 More derived tristichopterids (Eusthenopteron foordi, Eusthenopteron wenjukowi and 43 44 45 Platycephalichthys bischoffi) are in age and the most derived clade (from node 7) is 46 47 Famennian in age. 48 49 Concerning the origin of the Tristichopteridae whole group (except Tristichopterus 50 51 52 alatus, which is in the basal polytomy), node 11 is unambiguously reconstructed with a 53 54 Western North American origin. This Laurussian origin agrees with Ahlberg and Johanson 55 56 (1997) and disagrees with Young (2008) who argued for a Gondwanan origin of the whole 57 58 group. This Western North American origin is mainly due to the basal position of Tinirau 59 60

22 Society of Vertebrate Paleontology Page 23 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 clackae, which comes from this area. This is also due to the fact that Tristichopterus alatus 4 5 6 ends up in the basal polytomy. If it instead occupied a resolved and most basal position 7 8 within Tristichopteridae (as in Clément et al., 2009; Ahlberg and Johanson, 1997), the origin 9 10 of the entire group would be Western European. 11 12 The origin of node 10 is reconstructed with a Gondwanan origin (Antarctica). This is 13 14 15 explained by the basal position of Notorhizodon mackelveyi. According to these results, and 16 17 as previously proposed by Johanson and Ahlberg (2001), tristichopterids would have already 18 19 achieved a worldwide distribution as early as the Givetian. Our biogeographic analysis also 20 21 22 proposes dispersal from Gondwana (Antarctica) to Laurussia (Eastern Europe) during the 23 24 Givetian–Frasnian (node 10 to 9). 25 26 During the Late Devonian Period, our results support an expansion of tristichopterids 27 28 29 from Eastern Europe into Western Europe (node 8 to node 7). Much of tristichopterid history 30 31 took place in Laurussia (nodes 1 and 5–9), first in Eastern Europe (nodes 9, 8, and 1) and 32 33 then in Western Europe (nodes 5–7), with the introgression of one taxon (Eusthenopteron 34 35 foordi) into Western Laurussia (Canada). 36 37 38 According to these results, while some tristichopterids remained in Europe during the 39 40 Famennian (Heddleichthys dalgleisiensis and Langlieria socqueti), dispersal events occurred 41 42 from Western Europe to Australia (one from node 6 to 2 and another from node 5 to 4) and at 43 44 45 least once from Australia to Eastern North America (H. lindae). Eusthenodon wangsjoi, with 46 47 its remarkable distribution, expanded out of Australia, in the Famennian, to reach Greenland, 48 49 Europe, and Antarctica. These hypothesized Famennian tristichopterid dispersal events are 50 51 52 difficult to reconcile with modern paleogeographic reconstructions. We might expect such 53 54 discrepancies to result from our still fragmentary understanding of both tristichopterid 55 56 diversity and species–scale geographic distributions (most tristichopterid species are only 57 58 known from a single geographic locality). 59 60

23 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 24 of 51

1 2 3 The Gondwana-Euramerica Devonian vertebrate interchange (or Great Devonian 4 5 6 Interchange) (e.g., Janvier and Villarroel, 2000; Janvier and Clément, 2005; Young, 2008) is 7 8 a proposed southward dispersion of reputedly Euramerican endemics (, 9 10 Strepsodus and Asterolepis) in the Frasnian followed by a northward dispersion of 11 12 Gondwanan endemics (Groenlandaspis, phyllolepids, rhizodontids and megalichthyids) in 13 14 15 the Famennian. This Frasnian–Famennian faunal interchange, at least in the southward 16 17 direction, has recently been weakened by Olive et al. (in review). The present study supports 18 19 several interchanges between Euramerica and Gondwana during the Middle–Late Devonian. 20 21 22 The first is a southward migration during the Givetian (node 11 to 10) and could correspond 23 24 to a time-extended southward migration of the Great Devonian Interchange. The second is a 25 26 Givetian–Frasnian northward migration (node 10 to 9) that contradicts the Interchange, and 27 28 29 then there are several Famennian migrations, mainly in the southward direction. Although 30 31 our results are preliminary, they lend some support to tristichopterid distributional patterns 32 33 that are more complex than the proposed Great Devonian Interchange. 34 35 36 37 38 ACKNOWLEDGMENTS 39 40 41 42 We thank A. Impens and S. Berton (IRSNB) for preparation of part of the described 43 44 45 herein, and T. Hubin (IRSNB) for photographs. We are grateful to C. Cousin and A. Folie 46 47 (IRSNB) for curatorial support; to F. Solé (IRSNB) and K. Le Verger (MNHN) for their 48 49 support with RASP software; and to S. Ladevèze (MNHN) for her support with PAUP 50 51 52 software. We also thank D. Snitting (Uppsala University) for sending us a copy of his thesis 53 54 (Snitting, 2008a). We gratefully acknowledge M. Garcia Sanz at the X-ray Tomography 55 56 Imagery Platform AST-RX of the UMS2700, MNHN, Paris, France (funded by MNHN, 57 58 CNRS, Institut de France, Région Ile-de-France) for the X-ray tomography scans. Thanks are 59 60

24 Society of Vertebrate Paleontology Page 25 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 also due to F. Goussard and D. Germain (UMR 7207 MNHN/CNRS/UPMC) for their help in 4 5 6 the 3D modelling process. We thank the Gesves local council staff for providing technical 7 8 support and excavation permission in Strud. We are indebted to the successive Strud field 9 10 teams, who helped us since 2004. Finally, we thank the editor and the reviewers who helped 11 12 to improve the quality of the manuscript. This study was partly funded by the Belgian Federal 13 14 15 Science Policy Office (doctoral fellowship to S. Olive). 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

25 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 26 of 51

1 2 3 LITERATURE CITED 4 5 6 7 8 Ahlberg, P. E., and Z. Johanson. 1997. Second tristichopterid (Sarcopterygii, 9 10 Osteolepiformes) from the Upper Devonian of , New South Wales, 11 12 13 Australia, and phylogeny of the Tristichopteridae. Journal of Vertebrate Paleontology 14 15 17:653–673. 16 17 Ahlberg, P. E., and Z. Johanson. 1998. Osteolepiforms and the ancestry of tetrapods. 18 19 395:792–793. 20 21 22 Ahlberg, P. E., Z. Johanson, and E. B. Daeschler. 2001. The Late Devonian 23 24 (Sarcopterygii, Dipnoi) from Australia and North America, and its 25 26 biogeographical implications. Journal of Vertebrate Paleontology 21:1–12. 27 28 29 Andrews, S. M., and T. S. Westoll. 1970. The postcranial skeleton of Eusthenopteron foordi 30 31 Whiteaves. Transactions of the Royal Society of Edinburgh: Earth Sciences 68:207– 32 33 329. 34 35 36 Berg, L. S. 1937. A classification of -like vertebrates. Bulletin of the Academy of 37 38 Sciences of USSR, of Mathematics and Natural Sciences, Biological Series 39 40 1237–1277. 41 42 Béthune, P. de. 1954. Carte Géologique de Belgique (échelle 1/500.000) ; Atlas de Belgique. 43 44 45 Académie Royale de Belgique. 46 47 Borgen, U. J., and H. A. Nakrem. 2016. Morphology, phylogeny and of 48 49 osteolepiform fish. Fossils and Strata 61:1–514. 50 51 52 Clément, G. 2002. Large Tristichopteridae (Sarcopterygii, Tetrapodomorpha) from the Late 53 54 Famennian Evieux Formation of Belgium. Palaeontology 45:577–593. 55 56 57 58 59 60

26 Society of Vertebrate Paleontology Page 27 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 Clément, G., and C. A. Boisvert. 2006. Lohest’s true and false ‘Devonian ’: 4 5 6 evidence for the rhynchodipterid lungfish Soederberghia in the Famennian of 7 8 Belgium. Journal of Vertebrate Paleontology 26:276–283. 9 10 Clément, G., D. Snitting, and P. E. Ahlberg. 2009. A new tristichopterid (Sarcopterygii, 11 12 Tetrapodomorpha) from the Upper Famennian Evieux Formation (Upper Devonian) 13 14 15 of Belgium. Palaeontology 52:823–836. 16 17 Clément, G., P. E. Ahlberg, A. Blieck, H. Blom, J. A. Clack, E. Poty, J. Thorez, and P. 18 19 Janvier. 2004. Palaeogeography: Devonian tetrapod from western Europe. Nature 20 21 22 427:412–413. 23 24 Cloutier, R., and P. E. Ahlberg. 1996. Morphology, characters, and the interrelationships of 25 26 basal sarcopterygians; pp. 445–480 in M. L. J. Stiassny, L. R. Parenti, and G. D. 27 28 29 Johnson, G (eds.), Interrelationships of . Academic Press, San diego, California. 30 31 Cloutier, R., and A. -M. Candilier. 1995. Palaeozoic vertebrates of northern France and 32 33 Belgium: Part III - Sarcopterygii (Devonian to Carboniferous). Geobios Mémoire 34 35 spécial 19:335–342. 36 37 38 Cope, E. D. 1887. Zittel's Manual of Palaeontology. The American Naturalist 21:1014–1020. 39 40 Daeschler, E. B., and J. P. Downs. 2018. New description and diagnosis of Hyneria lindae 41 42 (Sarcopterygii, Tristichopteridae) from the Upper Devonian Catskill Formation in 43 44 45 Pennsylvania, USA. Journal of Vertebrate Paleontology:e1448834. 46 47 Daeschler, E. B., and N. H. Shubin. 2007. New data on Hyneria lindae (Sarcopterygii, 48 49 Tristichopteridae) from the Late Devonian of Pennsylvania, USA. Journal of 50 51 52 Vertebrate Paleontology 27 (3, Supplement):65A. 53 54 Davesne, D., J. Mondéjar-Fernández, V. Hairapetian, M. Rücklin, J. Wendt, and G. Clément. 55 56 2015. A new large tetrapodomorph sarcopterygian from the Late Devonian of Iran. 57 58 Paläontologische Zeitschrift 89:661–667. 59 60

27 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 28 of 51

1 2 3 Denayer, J., C. Prestianni, P. Gueriau, S. Olive, and G. Clément. 2016. Stratigraphy and 4 5 6 depositional environments of the Late Famennian (Late Devonian) of Southern 7 8 Belgium and characterization of the Strud locality. Geological Magazine 153:112– 9 10 127. 11 12 Downs, J. P., E. B. Daeschler, A. M. Long, and N. H. Shubin. 2018. Eusthenopteron jenkinsi 13 14 15 sp. nov. (Sarcopterygii, Tristichopteridae) from the Upper Devonian of Nunavut, 16 17 Canada, and a review of Eusthenopteron taxonomy). Breviora 562:1–24. 18 19 Egerton, P. de M. G. 1861. Tristichopterus alatus; pp. 51-55 in Figures and Descriptions 20 21 22 Illustrative of British Organic Remains. Memoirs of the Geological Survey of the 23 24 United Kingdom 10. 25 26 Holland, T., and J. A. Long. 2009. On the phylogenetic position of Gogonasus andrewsae 27 28 29 Long 1985, within the Tetrapodomorpha. Acta Zoologica, Supplement 1, 90:285–96. 30 31 Janvier, P., and G. Clément. 2005. A new groenlandaspidid Arthrodire (Vertebrata: 32 33 ) from the Famennian of Belgium. Geologica Belgica 8:51–67 34 35 Janvier, P. and J. G. Maisey. 2010. The Devonian vertebrates of and their 36 37 38 biogeographical relationships. pp.431-459 in D. K. Elliott, J. G. Maisey, X. Yu and D. 39 40 Miao (eds.), Morphology, phylogeny and paleobiogeography of fishes. Verlag 41 42 Dr. Friedrich Pfeil. München, Germany. 43 44 45 Janvier, P., and C. Villarroel. 2000. Devonian vertebrates from Colombia. Palaeontology 46 47 43:729–763. 48 49 Jarvik, E. 1944a. On the dermal bones, sensory canals and pit-lines of the skull in 50 51 52 Eusthenopteron foordi Whiteaves, with some remarks on E. save-soderberghi Jarvik. 53 54 Kungliga Svenska VetenskapsAkademiens Handlingar 21:1–48. 55 56 57 58 59 60

28 Society of Vertebrate Paleontology Page 29 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 Jarvik, E. 1944b. On the exoskeletal shoulder-girdle of teleostomian fishes, with special 4 5 6 reference to Eusthenopteron foordi Whiteaves. Kungliga Svenska 7 8 Vetenskapsakademiens Handlingar 21:1–32. 9 10 Jarvik, E. 1952. On the fish-like tail in the ichthyostegid stegocephalians with descriptions of 11 12 a new stegocephalian and a new crossopterygian from the Upper Devonian of East 13 14 15 Greenland. Meddelelser om Grønland 114:1–90. 16 17 Jarvik, E. 1985. Devonian osteolepiform fishes from East Greenland. Meddelelser om 18 19 Grønland, Geoscience 13:1–52. 20 21 22 Johanson, Z. 2004. Late Devonian sarcopterygian fishes from eastern Gondwana (Australia 23 24 and Antarctica) and their importance in phylogeny and biogeography. pp. 287–308 in 25 26 G. Arratia, M. V. H. Wilson, and R. Cloutier (eds.), Recent Advances in the Origin 27 28 29 and Early Radiation of Vertebrates. Verlag Dr. Friedrich Pfeil. München, Germany. 30 31 Johanson, Z., and P. E. Ahlberg. 1997. A new tristichopterid (Osteolepiformes: 32 33 Sarcopterygii) from the Mandagery Sandstone (Late Devonian, Famennian) near 34 35 Canowindra, NSW, Australia. Transactions of the Royal Society of Edinburgh: Earth 36 37 38 Sciences 88:39–68. 39 40 Johanson, Z., and P. E. Ahlberg. 2001. Devonian rhizodontids and tristichopterids 41 42 (Sarcopterygii; Tetrapodomorpha) from East Gondwana. Transactions of the Royal 43 44 45 Society of Edinburgh: Earth Sciences 92:43–74. 46 47 Lehman, J. -P. 1977. Sur la présence d’un ostéolépiforme dans le Dévonien supérieur du 48 49 Tafilalet. Comptes Rendus de l’Académie des Sciences, Série D, Sciences naturelles 50 51 52 285:151–153. 53 54 Lelièvre, H., and P. Janvier. 1986. L'eusthénoptéridé (, Sarcopterygii) du 55 56 Famennien (Dévonien supérieur) du Tafilalet (Maroc): nouvelle description. Bulletin 57 58 59 60

29 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 30 of 51

1 2 3 du Muséum National d'Histoire Naturelle, Section C, Sciences de la terre, 4 5 6 paléontologie, géologie, minéralogie 3:351–365. 7 8 Lelièvre, H., P. Janvier, and A. Blieck. 1993. -Devonian vertebrate biostratigraphy of 9 10 Western Gondwana and related terranes (South America, Africa, Armorica-Bohemia, 11 12 Middle East); pp. 139–173 in J. A. Long (ed.), Palaeozoic Vertebrate Biostratigraphy 13 14 15 and Biogeography, Belhaven Press, London. 16 17 Leriche, M. 1931. Les poissons famenniens de la Belgique. Les faciès du Famennien dans la 18 19 région gallo-belge. Les relations entre les formations marines et les formations 20 21 22 continentales du Dévonien supérieur sur la bordure méridionale du Continent Nord- 23 24 Atlantique. Mémoires de l’Académie Royale de Belgique Classe des Sciences 10:1–72. 25 26 Lohest, M. 1888. Recherches sur les poissons des terrains paléozoïques de Belgique. 27 28 29 Poissons des Psammites du Condroz, Famennien supérieur. Annales de la Société 30 31 Géologique de Belgique Mémoire 15:112–203. 32 33 Long, J. A. 1985a. A new osteolepidid fish from the Upper Devonian of 34 35 Western Australia. Records of the Western Australian Museum 12:361–377. 36 37 38 Long, J. A. 1985b. The structure and relationships of a new osteolepiform fish from the Late 39 40 Devonian of Victoria, Australia. Alcheringa 9:1–22. 41 42 Long, J. A., G. C. Young, T. Holland, T. J. Senden, and E. M. G. Fitzgerald. 2006. An 43 44 45 exceptional Devonian fish from Australia sheds light on tetrapod evolution. Nature 46 47 444:199–202. 48 49 Olive, S. 2015. Devonian antiarch placoderms from Belgium revisited. Acta Palaeontologica 50 51 52 Polonica 60:711–731. 53 54 Olive, S., G. Clément, E. B. Daeschler, and V. Dupret. 2015b. Characterization of the 55 56 placoderm () assemblage from the tetrapod-bearing locality of Strud 57 58 (Belgium, Upper Famennian). Palaeontology 58:981–1002. 59 60

30 Society of Vertebrate Paleontology Page 31 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 Olive, S., G. Clément, E. B. Daeschler, and V. Dupret. 2016b. Placoderm assemblage from 4 5 6 the tetrapod-bearing locality of Strud (Belgium, Upper Famennian) provides evidence 7 8 for a fish nursery. PLoS ONE 11:e0161540. 9 10 Olive, S., P. E. Ahlberg, V. N. Pernègre, E. Poty, E. Steurbaut, and G. Clément. 2016a. New 11 12 discoveries of tetrapods (ichthyostegid-like and whatcheeriid-like) in the Famennian 13 14 15 (Late Devonian) localities of Strud and Becco (Belgium). Palaeontology 59:827–840. 16 17 Olive, S., G. Clément, J. Denayer, C. Derycke, V. Dupret, P. Gerrienne, P. Gueriau, J. –M. 18 19 Marion, B. Mottequin, and C. Prestianni. 2015a. Flora and fauna from a new 20 21 22 Famennian (Upper Devonian) locality at Becco, eastern Belgium. Geologica Belgica 23 24 18:92–101. 25 26 27 Olive, S., A. Pradel, C. Martinez-Perez, J. Lamsdell, P. Gueriau, N. Rabet, P. Duranleau- 28 29 Gagnon, P. Janvier, A. L. Cardenas, P. A. Zapata-Ramirez, and H. Botella. In review. 30 31 New insights into Devonian vertebrates from Colombia and associate fauna. Journal 32 33 34 of Vertebrate Paleontology. 35 36 Romer, A. S. 1955. Herpetichthyes, Amphibioidei, Choanichthyes or Sarcopterygii? Nature 37 38 39 176:126. 40 41 Ronquist, F., and J. P. Huelsenbeck. 2003. MrBayes3: Bayesian phylogenetic inference 42 43 undermixed models. Bioinformatics 19:1572–1574. 44 45 46 Schultze, H. -P., and J. W. Reed. 2012. A tristichopterid sarcopterygian fish from the upper 47 48 Middle Devonian of Nevada. Historical Biology 24:425–440. 49 50 Scotese, C. R. 2014. Atlas of Devonian paleogeographic maps, PALEOMAP atlas for 51 52 53 ArcGIS, 4, the Late Paleozoic, Maps 65-72, Mollweide Projection, PALEOMAP 54 55 Project, Evanston, Illinois. 56 57 58 59 60

31 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 32 of 51

1 2 3 Snitting, D. 2008a. Morphology, taxonomy and interrelationships of tristichopterid fishes 4 5 6 (Sarcopterygii, Tetrapodomorpha). Ph.D. dissertation, Acta Universitatis Upsaliensis, 7 8 Uppsala, Sweden, 54 pp. 9 10 Snitting, D. 2008b. A redescription of the anatomy of the Late Devonian Spodichthys buetleri 11 12 Jarvik, 1985 (Sarcopterygii, Tetrapodomorpha) from East Greenland. Journal of 13 14 15 Vertebrate Paleontology 28:637–655. 16 17 Snitting, D. 2009. Heddleichthys - a new tristichopterid from the Dura Den Formation, 18 19 Midland Valley, Scotland (Famennian, Late Devonian). Acta Zoologica 90(1, 20 21 22 Supplement):273–284. 23 24 Swartz, B. 2012. A marine stem-tetrapod from the Devonian of western North America. 25 26 PLoS ONE 7:e33683. 27 28 29 Swofford, D. L. 2002. PAUP*: phylogenetic analysis using parcimony (and other methods). 30 31 Version 4.0a62. http://phylosolutions.com/paup-test/ 32 33 Thorez, J., and R. Dreesen. 1986. A model of regressive depositional system around the Old 34 35 Red Continent as exemplified by a field trip in the Upper Famennian ‘Psammites du 36 37 38 Condroz’ in Belgium. Annales de la Société géologique de Belgique 109:285–323. 39 40 Thorez, J., R. Dreesen, and M. Streel. 2006. Famennian; pp. 27–45 in L. Dejonghe (ed.), 41 42 Current status of chronostratigraphic units named from Belgium and adjacent areas. 43 44 45 Geologica Belgica 9. 46 47 Vorobyeva, E. I. 1959. A new crossopterygian fish genus Platycephalichthys from the Upper 48 49 Devonian of the Lovat . Paleontologicheskii Zhurnal, Acd. Sci. U.S.S.R. 3:95– 50 51 52 106. [Russian] 53 54 Vorobyeva, E. I. 1962. Rhizodontid crossopterygians of the main Devonian field of the 55 56 USSR. Trudy Paleontologicheskogo Instituta, Akademii Nauk SSSR 94:1–139. 57 58 [Russian] 59 60

32 Society of Vertebrate Paleontology Page 33 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 Young, G. C. 2008. Relationships of tristichopterids (osteolepiform lobe-finned fishes) from 4 5 6 the Middle–Late Devonian of East Gondwana. Alcheringa 32:321–336. 7 8 Young, G. C., J. A. Long, and A. Ritchie. 1992. Crossopterygian fishes from the Devonian of 9 10 Antarctica: systematics, relationships and biogeographic significance. Records of the 11 12 Australian Museum Supplement 14:1–77. 13 14 15 Young, B., R. L. Dunstone, T. J. Senden, and G. C. Young. 2013. A Gigantic Sarcopterygian 16 17 (Tetrapodomorph Lobe-Finned Fish) from the Upper Devonian of Gondwana (Eden, 18 19 New South Wales, Australia). PLoS ONE 8:e53871. 20 21 22 Yu, Y., A. J. Harris, C. Blair, and X. J. He. 2015. RASP (Reconstruct Ancestral State in 23 24 Phylogenies): a tool for historical biogeography. Molecular Phylogenetics and 25 26 Evolution 87:46–49. 27 28 29 Zhu, M., and P. E. Ahlberg. 2004. The origin of the internal nostril of tetrapods. Nature 30 31 432:94–97. 32 33 Ziegler, P. A. 1990. Geological atlas of Western and Central Europe, Second edition. Shell 34 35 Internationale Petroleum Maatschappij B.V., and Geological Society of London, 36 37 38 Avon, 239 pp. 39 40 41 42 43 44 45 Submitted April 5th, 2019; revisions received Month DD, YYYY; accepted Month DD, 46 47 YYYY 48 49 50 51 52 53 54 55 56 57 58 59 60

33 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 34 of 51

1 2 3 Figure Captions 4 5 6 7 8 FIGURE 1. Geographical and geological localizations of Belgian tristichopterid localities 9 10 including Strud (Namur Province). A, Geographic localities; B, Geological map of Southern 11 12 Belgium, modified from Denayer et al. (2016), redrawn from Béthune (1954); C, schematic 13 14 15 block diagram of the Strud channel with the location of tristichopterid remains, modified 16 17 from Denayer et al (2016). [planned for page width] 18 19 20 21 22 FIGURE 2. cf. Langlieria socqueti, Strud, Belgium, late Famennian. A, left squamosal, 23 24 internal view, IRSNB P 9912, photograph; B, left squamosal, imprint of the external surface 25 26 (bone removed), IRSNB P 9912, photograph; C, left squamosal, external view, IRSNB P 27 28 29 9912, interpretative drawing; D, right mandible, external view, IRSNB P 9913, photograph; 30 31 E, right mandible, external view, IRSNB P 9913, interpretative drawing; F, right mandible, 32 33 external view, IRSNB P 9913, close-up photograph on the symphysis. Scale bars equal 1 cm. 34 35 [planned for page width] 36 37 38 39 40 FIGURE 3. cf. Eusthenodon wangsjoi, Strud, Belgium, Upper Famennian. A, incomplete left 41 42 parietal, external view, IRSNB P 9910, photograph; B, incomplete left parietal, external 43 44 45 view, IRSNB P 9910, interpretative drawing; C–D, left supratemporal, external and visceral 46 47 views, IRSNB P 9914, photographs; E–F, left supratemporal, external and visceral views, 48 49 IRSNB P 9914, interpretative drawings; G, left postorbital, external view, IRSNB P 9915, 50 51 52 photograph; H, left postorbital, external view, IRSNB P 9915, interpretative drawing; I–J, 53 54 left preopercular, external and visceral views, IRSNB P 9916, photographs; K, left opercular, 55 56 external view, IRSNB P 9917, photograph; L, left opercular, external view, IRSNB P 9917, 57 58 59 60

34 Society of Vertebrate Paleontology Page 35 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 interpretative drawing. Black arrow points anteriorly. Scale bars: A–H equal 1 cm, I–L equal 4 5 6 2 cm. [planned for page width] 7 8 9 10 FIGURE 4. cf. Eusthenodon wangsjoi, Strud, Belgium, late Famennian. A, right dentary, 11 12 external view, IRSNB P 9918, 3D reconstruction; B, right dentary, visceral view, IRSNB P 13 14 15 9918, 3D reconstruction; C, right mandible, external view, IRSNB P 9919, 3D 16 17 reconstruction; D, right mandible, visceral view, IRSNB P 9919, 3D reconstruction; E, 18 19 coronoids, external view, IRSNB P 9919, 3D reconstruction; F, coronoids, apical view, 20 21 22 IRSNB P 9919, 3D reconstruction. Left white arrow indicates putative suture between the 23 24 infradentaries 3 and 4, right white arrow indicates pit line of infradentary 2. Black arrow 25 26 points anteriorly. Scale bars: A–B, E–F equal 1 cm, C–D equal 2 cm. [planned for page 27 28 29 width] 30 31 32 33 FIGURE 5. cf. Tristichopteridae indet., Strud, Belgium, Upper Famennian. A, right 34 35 preopercular, external view, IRSNB P 9920, photograph; B, left subopercular, visceral view, 36 37 38 IRSNB P 9921, photograph; C, left subopercular, visceral view, IRSNB P 9922, photograph; 39 40 D, left subopercular, visceral view, IRSNB P 9923, photograph; E, right palatoquadrate, 41 42 visceral view, IRSNB P 9911, photograph; F, right palatoquadrate, visceral view, IRSNB P 43 44 45 9911, interpretative drawing. Black arrow points anteriorly. Scale bars equal 1 cm. [planned 46 47 for page width] 48 49 50 51 52 FIGURE 6. cf. Tristichopteridae indet., Strud, Belgium, Upper Famennian. A, right 53 54 cleithrum, visceral view, IRSNB P 9924, photograph; B, right cleithrum, imprint of the 55 56 external view (bone removed), IRSNB P 9924, photograph; C–D, right cleithrum, external 57 58 and visceral views, IRSNB P 9925, photographs; E, incomplete right cleithrum, visceral 59 60

35 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 36 of 51

1 2 3 view, IRSNB P 9926, photograph; F, incomplete right cleithrum, visceral view, IRSNB P 4 5 6 9927, photograph; G, isolated fang, labial or lingual view, IRSNB P 9928, photograph; H, 7 8 isolated fang, labial or lingual view, IRSNB P 9929, photograph; I, isolated scale, visceral 9 10 view, IRSNB P 9930, photograph; J, isolated scale, visceral view, IRSNB P 9931, 11 12 photograph; K, isolated scale, visceral view, IRSNB P 9932, photograph; L, isolated scale, 13 14 15 visceral view, IRSNB P 9933, photograph. Scale bars equal 1 cm. [planned for page width] 16 17 18 19 FIGURE 7. Strict consensus tree showing phylogenetic relationships of the Tristichopteridae, 20 21 22 with maximum parsimony biogeographic reconstruction result using Bayesian Binary 23 24 MCMC analysis mapped on. The number in the center of each pie chart corresponds to the 25 26 node number. The pie chart over each node represents the probability that the clade 27 28 29 originated from each continental area. The numbers on the top of each pie chart correspond to 30 31 the Bremer index (in italics) and to the minimal and maximal number of synapomorphies at 32 33 each node (in bold). Some character state changings have been reported on the branches. GIV 34 35 for Givetian, FRAS for Frasnian and FAM for Famennian. [planned for page width] 36 37 38 39 40 SUPPLEMENTAL INFORMATION 1. List of characters considered in the phylogenetic 41 42 analysis. 43 44 45 46 47 SUPPLEMENTAL INFORMATION 2. Matrix of characters considered in the phylogenetic 48 49 analysis. 50 51 52 53 54 SUPPLEMENTAL INFORMATION 3. Details of the phylogenetic analysis. 55 56 57 58 SUPPLEMENTAL INFORMATION 4. Taxon geographical distribution. 59 60

36 Society of Vertebrate Paleontology Page 37 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 4 5 6 SUPPLEMENTAL INFORMATION 5. Results of the maximum parsimony biogeographic 7 8 reconstructions using Bayesian Binary MCMC analysis. 9 10 11 12 SUPPLEMENTARY INFORMATION 6: 3D PDF of a right dentary of Eusthenodon 13 14 15 wangsjoi. Strud, Belgium, late Famennian. IRSNB P 9918. Click to activate in the PDF 16 17 version. 18 19 20 21 22 SUPPLEMENTARY INFORMATION 7: 3D PDF of a right mandible of Eusthenodon 23 24 wangsjoi. Strud, Belgium, late Famennian. IRSNB P 9919. Click to activate in the PDF 25 26 version. 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

37 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 38 of 51

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 FIGURE 1. Geographical and geological localizations of Belgian tristichopterid localities including Strud 37 (Namur Province). A, Geographic localities; B, Geological map of Southern Belgium, modified from Denayer et al. (2016), redrawn from Béthune (1954); C, schematic block diagram of the Strud channel with the 38 location of tristichopterid remains, modified from Denayer et al (2016). [planned for page width] 39 40 180x163mm (300 x 300 DPI) 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Society of Vertebrate Paleontology Page 39 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 FIGURE 1. Geographical and geological localizations of Belgian tristichopterid localities including Strud 37 (Namur Province). A, Geographic localities; B, Geological map of Southern Belgium, modified from Denayer et al. (2016), redrawn from Béthune (1954); C, schematic block diagram of the Strud channel with the 38 location of tristichopterid remains, modified from Denayer et al (2016). [planned for page width] 39 40 180x163mm (300 x 300 DPI) 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 40 of 51

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 FIGURE 2. cf. Langlieria socqueti, Strud, Belgium, late Famennian. A, left squamosal, internal view, IRSNB P 39 9912, photograph; B, left squamosal, imprint of the external surface (bone removed), IRSNB P 9912, photograph; C, left squamosal, external view, IRSNB P 9912, interpretative drawing; D, right mandible, 40 external view, IRSNB P 9913, photograph; E, right mandible, external view, IRSNB P 9913, interpretative 41 drawing; F, right mandible, external view, IRSNB P 9913, close-up photograph on the symphysis. Scale bars 42 equal 1 cm. [planned for page width] 43 44 181x176mm (300 x 300 DPI) 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Society of Vertebrate Paleontology Page 41 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 FIGURE 2. cf. Langlieria socqueti, Strud, Belgium, late Famennian. A, left squamosal, internal view, IRSNB P 39 9912, photograph; B, left squamosal, imprint of the external surface (bone removed), IRSNB P 9912, photograph; C, left squamosal, external view, IRSNB P 9912, interpretative drawing; D, right mandible, 40 external view, IRSNB P 9913, photograph; E, right mandible, external view, IRSNB P 9913, interpretative 41 drawing; F, right mandible, external view, IRSNB P 9913, close-up photograph on the symphysis. Scale bars 42 equal 1 cm. [planned for page width] 43 44 181x176mm (300 x 300 DPI) 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 42 of 51

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 FIGURE 3. cf. Eusthenodon wangsjoi, Strud, Belgium, Upper Famennian. A, incomplete left parietal, external 46 view, IRSNB P 9910, photograph; B, incomplete left parietal, external view, IRSNB P 9910, interpretative 47 drawing; C–D, left supratemporal, external and visceral views, IRSNB P 9914, photographs; E–F, left supratemporal, external and visceral views, IRSNB P 9914, interpretative drawings; G, left postorbital, 48 external view, IRSNB P 9915, photograph; H, left postorbital, external view, IRSNB P 9915, interpretative 49 drawing; I–J, left preopercular, external and visceral views, IRSNB P 9916, photographs; K, left opercular, 50 external view, IRSNB P 9917, photograph; L, left opercular, external view, IRSNB P 9917, interpretative 51 drawing. Black arrow points anteriorly. Scale bars: A–H equal 1 cm, I–L equal 2 cm. [planned for page 52 width] 53 184x260mm (300 x 300 DPI) 54 55 56 57 58 59 60 Society of Vertebrate Paleontology Page 43 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 FIGURE 3. cf. Eusthenodon wangsjoi, Strud, Belgium, Upper Famennian. A, incomplete left parietal, external 46 view, IRSNB P 9910, photograph; B, incomplete left parietal, external view, IRSNB P 9910, interpretative 47 drawing; C–D, left supratemporal, external and visceral views, IRSNB P 9914, photographs; E–F, left supratemporal, external and visceral views, IRSNB P 9914, interpretative drawings; G, left postorbital, 48 external view, IRSNB P 9915, photograph; H, left postorbital, external view, IRSNB P 9915, interpretative 49 drawing; I–J, left preopercular, external and visceral views, IRSNB P 9916, photographs; K, left opercular, 50 external view, IRSNB P 9917, photograph; L, left opercular, external view, IRSNB P 9917, interpretative 51 drawing. Black arrow points anteriorly. Scale bars: A–H equal 1 cm, I–L equal 2 cm. [planned for page 52 width] 53 184x260mm (300 x 300 DPI) 54 55 56 57 58 59 60 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 44 of 51

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 FIGURE 4. cf. Eusthenodon wangsjoi, Strud, Belgium, late Famennian. A, right dentary, external view, 46 IRSNB P 9918, 3D reconstruction; B, right dentary, visceral view, IRSNB P 9918, 3D reconstruction; C, right 47 mandible, external view, IRSNB P 9919, 3D reconstruction; D, right mandible, visceral view, IRSNB P 9919, 3D reconstruction; E, coronoids, external view, IRSNB P 9919, 3D reconstruction; F, coronoids, apical view, 48 IRSNB P 9919, 3D reconstruction. Left white arrow indicates putative suture between the infradentaries 3 49 and 4, right white arrow indicates pit line of infradentary 2. Black arrow points anteriorly. Scale bars: A–B, 50 E–F equal 1 cm, C–D equal 2 cm. [planned for page width] 51 52 181x257mm (300 x 300 DPI) 53 54 55 56 57 58 59 60 Society of Vertebrate Paleontology Page 45 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 FIGURE 4. cf. Eusthenodon wangsjoi, Strud, Belgium, late Famennian. A, right dentary, external view, 46 IRSNB P 9918, 3D reconstruction; B, right dentary, visceral view, IRSNB P 9918, 3D reconstruction; C, right 47 mandible, external view, IRSNB P 9919, 3D reconstruction; D, right mandible, visceral view, IRSNB P 9919, 3D reconstruction; E, coronoids, external view, IRSNB P 9919, 3D reconstruction; F, coronoids, apical view, 48 IRSNB P 9919, 3D reconstruction. Left white arrow indicates putative suture between the infradentaries 3 49 and 4, right white arrow indicates pit line of infradentary 2. Black arrow points anteriorly. Scale bars: A–B, 50 E–F equal 1 cm, C–D equal 2 cm. [planned for page width] 51 52 181x257mm (300 x 300 DPI) 53 54 55 56 57 58 59 60 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 46 of 51

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 FIGURE 5. cf. Tristichopteridae indet., Strud, Belgium, Upper Famennian. A, right preopercular, external 45 view, IRSNB P 9920, photograph; B, left subopercular, visceral view, IRSNB P 9921, photograph; C, left 46 subopercular, visceral view, IRSNB P 9922, photograph; D, left subopercular, visceral view, IRSNB P 9923, 47 photograph; E, right palatoquadrate, visceral view, IRSNB P 9911, photograph; F, right palatoquadrate, 48 visceral view, IRSNB P 9911, interpretative drawing. Black arrow points anteriorly. Scale bars equal 1 cm. 49 [planned for page width] 50 181x213mm (300 x 300 DPI) 51 52 53 54 55 56 57 58 59 60 Society of Vertebrate Paleontology Page 47 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 FIGURE 5. cf. Tristichopteridae indet., Strud, Belgium, Upper Famennian. A, right preopercular, external 45 view, IRSNB P 9920, photograph; B, left subopercular, visceral view, IRSNB P 9921, photograph; C, left 46 subopercular, visceral view, IRSNB P 9922, photograph; D, left subopercular, visceral view, IRSNB P 9923, 47 photograph; E, right palatoquadrate, visceral view, IRSNB P 9911, photograph; F, right palatoquadrate, 48 visceral view, IRSNB P 9911, interpretative drawing. Black arrow points anteriorly. Scale bars equal 1 cm. 49 [planned for page width] 50 181x213mm (300 x 300 DPI) 51 52 53 54 55 56 57 58 59 60 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 48 of 51

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 FIGURE 6. cf. Tristichopteridae indet., Strud, Belgium, Upper Famennian. A, right cleithrum, visceral view, 39 IRSNB P 9924, photograph; B, right cleithrum, imprint of the external view (bone removed), IRSNB P 9924, 40 photograph; C–D, right cleithrum, external and visceral views, IRSNB P 9925, photographs; E, incomplete 41 right cleithrum, visceral view, IRSNB P 9926, photograph; F, incomplete right cleithrum, visceral view, IRSNB P 9927, photograph; G, isolated fang, labial or lingual view, IRSNB P 9928, photograph; H, isolated 42 fang, labial or lingual view, IRSNB P 9929, photograph; I, isolated scale, visceral view, IRSNB P 9930, 43 photograph; J, isolated scale, visceral view, IRSNB P 9931, photograph; K, isolated scale, visceral view, 44 IRSNB P 9932, photograph; L, isolated scale, visceral view, IRSNB P 9933, photograph. Scale bars equal 1 45 cm. [planned for page width] 46 47 181x178mm (300 x 300 DPI) 48 49 50 51 52 53 54 55 56 57 58 59 60 Society of Vertebrate Paleontology Page 49 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 FIGURE 6. cf. Tristichopteridae indet., Strud, Belgium, Upper Famennian. A, right cleithrum, visceral view, 39 IRSNB P 9924, photograph; B, right cleithrum, imprint of the external view (bone removed), IRSNB P 9924, 40 photograph; C–D, right cleithrum, external and visceral views, IRSNB P 9925, photographs; E, incomplete 41 right cleithrum, visceral view, IRSNB P 9926, photograph; F, incomplete right cleithrum, visceral view, IRSNB P 9927, photograph; G, isolated fang, labial or lingual view, IRSNB P 9928, photograph; H, isolated 42 fang, labial or lingual view, IRSNB P 9929, photograph; I, isolated scale, visceral view, IRSNB P 9930, 43 photograph; J, isolated scale, visceral view, IRSNB P 9931, photograph; K, isolated scale, visceral view, 44 IRSNB P 9932, photograph; L, isolated scale, visceral view, IRSNB P 9933, photograph. Scale bars equal 1 45 cm. [planned for page width] 46 47 181x178mm (300 x 300 DPI) 48 49 50 51 52 53 54 55 56 57 58 59 60 Society of Vertebrate Paleontology Journal of Vertebrate Paleontology: For Review Only Page 50 of 51

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 FIGURE 7. Strict consensus tree showing phylogenetic relationships of the Tristichopteridae, with maximum parsimony biogeographic reconstruction result using Bayesian Binary MCMC analysis mapped on. The 42 number in the center of each pie chart corresponds to the node number. The pie chart over each node 43 represents the probability that the clade originated from each continental area. The numbers on the top of 44 each pie chart correspond to the Bremer index (in italics) and to the minimal and maximal number of 45 synapomorphies at each node (in bold). Some character state changings have been reported on the 46 branches. GIV for Givetian, FRAS for Frasnian and FAM for Famennian. [planned for page width] 47 182x193mm (300 x 300 DPI) 48 49 50 51 52 53 54 55 56 57 58 59 60 Society of Vertebrate Paleontology Page 51 of 51 Journal of Vertebrate Paleontology: For Review Only

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 FIGURE 7. Strict consensus tree showing phylogenetic relationships of the Tristichopteridae, with maximum parsimony biogeographic reconstruction result using Bayesian Binary MCMC analysis mapped on. The 42 number in the center of each pie chart corresponds to the node number. The pie chart over each node 43 represents the probability that the clade originated from each continental area. The numbers on the top of 44 each pie chart correspond to the Bremer index (in italics) and to the minimal and maximal number of 45 synapomorphies at each node (in bold). Some character state changings have been reported on the 46 branches. GIV for Givetian, FRAS for Frasnian and FAM for Famennian. [planned for page width] 47 182x193mm (300 x 300 DPI) 48 49 50 51 52 53 54 55 56 57 58 59 60 Society of Vertebrate Paleontology